Complex systems are complex, evolutionary, and dynamical system. One general method to predict such systems is use the previous and most recently behavior of a system to predict its future changes. The main advantage of this method is the ability to predict the behavior of systems without analytical prediction rules. In this situation, decision makers are often presented with several competing forecasts produced by different forecasting methods. A decision maker who needs a predict could choose a combined forecast that is generally more precise than any of the individual forecasts, for the combined forecast gets more information into consideration and the preciseness of the combined forecast improves as more methods are included in the combination. This article proposes a new forecast combination strategy, by using support vector machines that improve the forecasting capability of the model. Finally the results of using this method on two sample datasets are presented and the superiority of this method is demonstrated.